fix: Bugfix and add retry logic to dynamodb as onlinestore

[iamcodingcat]

What this PR does / why we need it:

A few weeks ago, I made PR and merge it for integrating dynamodb as online store in Golang feature server. And before applying it in my production, I run load test with locust in EKS cluster. During load testing, One bug is found and I found the one thing that I will enhance when requesting to dynamodb.

Which issue(s) this PR fixes:

I made two features in Golang feature server. the explanation for them is below.

1. Bugfix: Before revising the commit, unprocessedEntityIds variable maintains processed entity keys. But this variable is globally defined outside feature view loop. So, if some entity keys doesn't exist in A feature view, then delete them from unprocessedEntityIds. Because of this, although the entity keys exist in B feature view, Golang server requests to dynamodb not including the entity keys. At the result, the below bug occurs nil derefrence. The stack trace is below.

{"level":"error",
"stack_trace":"goroutine 121881 [running]:\ngithub.com/feast-dev/feast/go/internal/feast/server.logStackTrace()\n\t
/feast/go/internal/feast/server/http_server.go:304 +0x64\n
github.com/feast-dev/feast/go/internal/feast/server.(*httpServer).Serve.recoverMiddleware.func1.1()\n\t
/feast/go/internal/feast/server/http_server.go:333 +0x98\n
panic({0x1a3fd20?, 0x2db8d60?})\n\t
/usr/local/go/src/runtime/panic.go:792 +0x124\n
github.com/feast-dev/feast/go/internal/feast/onlineserving.TransposeFeatureRowsIntoColumns({0x4001662008, 0x1f3, 0x40010820a0?}, 0x4001196580, {0x4001148fc0, 0x5, 0x400567e800?}, {0x1f5c680, 0x2f1ede0}, 0x1f4)\n\t
/feast/go/internal/feast/onlineserving/serving.go:388 +0x424\ngithub.com/feast-dev/feast/go/internal/feast.(*FeatureStore).GetOnlineFeatures(0x4000976fc0, {0x1f61e98, 0x40010820a0}, {0x4001100008, 0x2f, 0x47}, 0x0, 0x4003968db0, 0x4003968de0, 0x0)\n\t
/feast/go/internal/feast/featurestore.go:173 +0x5d0\n
github.com/feast-dev/feast/go/internal/feast/server.(*httpServer).getOnlineFeatures(0x40004c5f08, {0x1f5cc20, 0x400087e1c0}, 0x40008deb40)\n\t
/feast/go/internal/feast/server/http_server.go:201 +0x6a8\n
net/http.HandlerFunc.ServeHTTP(0x8080808080806a36?, {0x1f5cc20?, 0x400087e1c0?}, 0x2ef5480?)\n\t
/usr/local/go/src/net/http/server.go:2294 +0x38\ngithub.com/feast-dev/feast/go/internal/feast/server.(*httpServer).Serve.recoverMiddleware.func1({0x1f5cc20?, 0x400087e1c0?}, 0x4004653b28?)\n\t
/feast/go/internal/feast/server/http_server.go:338 +0x6c\n
net/http.HandlerFunc.ServeHTTP(0x400088b200?, {0x1f5cc20?, 0x400087e1c0?}, 0x4004653b10?)\n\t
/usr/local/go/src/net/http/server.go:2294 +0x38\n
net/http.(*ServeMux).ServeHTTP(0x10?, {0x1f5cc20, 0x400087e1c0}, 0x40008deb40)\n\t
/usr/local/go/src/net/http/server.go:2822 +0x1b4\n
net/http.serverHandler.ServeHTTP({0x1f53028?}, {0x1f5cc20?, 0x400087e1c0?}, 0x6?)\n\t
/usr/local/go/src/net/http/server.go:3301 +0xbc\n
net/http.(*conn).serve(0x400241a240, {0x1f61e60, 0x40008b1230})\n\t
/usr/local/go/src/net/http/server.go:2102 +0x52c\ncreated by net/http.(*Server).Serve in goroutine 1\n\t
/usr/local/go/src/net/http/server.go:3454 +0x3d8\n","time":"2025-06-27T07:36:01Z"
}

So, I move defining unprocessedEntityIds(in the new commit, I rename unprocessedEntityIdsFeatureView variable) in feature view loop.

2. Enhancement: I add dynamodb retrying related config and retry logic when request to dynamodb. Of course, the revision in dynamodb-side configuration(Read units per seconds, ...) will be needed but I think that client side retrty logic must be needed. So I add retry logic with exponential backoff and max retry config value is set to dynamodb online store config that can be defined in feature_store.yaml.

Misc

@shuchu @jyejare cc. @franciscojavierarceo

[ntkathole]

Looks good

[iamcodingcat]

@ntkathole Thanks for your fast review!